zarzycki
New Member
Worth posting in case someone else has seen this. Attempting to build F compset with SE dycore on Hopper using stock v1.2.2 release (cam5.3.1, IIRC).Have no problems building/running a test CESM configuration with FV (e.g, 0.9x0.125 grid). However, if I try to build something like...CASE_NAME=ne120_FC5_v122
CASEROOT=$SCRATCH/${CASE_NAME}
./create_newcase -res ne120_ne120 -compset FC5 -mach hopper -case ${CASEROOT}
./cesm_setup
./${CASE_NAME}.buildthe build hangs during the atm dycore build stage. It appears the hang occurs at or just after dyn_comp.F90, which would make sense (it is an dycore-specific file, and I would assume the only reason an FV compset would build and SE would not is due to issues in the dycore).This occurs regardless of playing with env_mach_specific to try multiple PGI versions (in addition to using multiple env_mach_specific files that have been recently updated after some Hopper upgrades).Another CESM user has mentioned that he has run into this issue before, although his solution has essentially been "if I log out and log in and rebuild, eventually it goes through). A) I have not been able to replicate this success and B) it doesn't seem like an effective fix, especially since I frequently rebuild with dev code.I am in contact with support at NERSC but they have not been able to nail down the issue as of yet. Also worth noting that I have run into the same problem with a CESM 1.4.X beta tag.Anyone else with more exp. with either Hopper/PGI/etc. than me have any suggestions?
CASEROOT=$SCRATCH/${CASE_NAME}
./create_newcase -res ne120_ne120 -compset FC5 -mach hopper -case ${CASEROOT}
./cesm_setup
./${CASE_NAME}.buildthe build hangs during the atm dycore build stage. It appears the hang occurs at or just after dyn_comp.F90, which would make sense (it is an dycore-specific file, and I would assume the only reason an FV compset would build and SE would not is due to issues in the dycore).This occurs regardless of playing with env_mach_specific to try multiple PGI versions (in addition to using multiple env_mach_specific files that have been recently updated after some Hopper upgrades).Another CESM user has mentioned that he has run into this issue before, although his solution has essentially been "if I log out and log in and rebuild, eventually it goes through). A) I have not been able to replicate this success and B) it doesn't seem like an effective fix, especially since I frequently rebuild with dev code.I am in contact with support at NERSC but they have not been able to nail down the issue as of yet. Also worth noting that I have run into the same problem with a CESM 1.4.X beta tag.Anyone else with more exp. with either Hopper/PGI/etc. than me have any suggestions?